New Features for Phased Array Techniques Inspections : Simulation and Experiments

Ultrasonic NDT techniques based on phased array technology are more and more applied in various industrial contexts, as they provide improved adaptability to different inspection configurations. Such techniques include conventional or advanced inspection modes, as electronic commutation, sectorial scanning, adaptive beam forming, multi-channels data reconstructions... All those features shall be performed over canonical or complex profile specimen. These techniques rely on the application of delay and amplitude laws to all, or to groups of, elements of an array. Within the CIVA software developed by CEA are included a set of tools based on modelling and developed specifically for phased array applications. Delay laws computation tools are the first of them. Indeed as soon as complex structures are dealt with (complex profile, non homogeneous and/or anisotropic materials...) the laws have to be optimised in order to compensate phase aberrations due to geometrical or material properties variations and this is done within CIVA by applying suitable modelling based algorithms. The second application of modelling concerns the simulation of the inspection by predicting either the beam transmitted beam in the part or the echoes arising from flaws. These simulations allow to evaluate the performances of methods and/or interpret experimental results. At last modelling based reconstruction algorithms can be applied to the experimental results in the aim of accurately locating the ultrasonic echoes inside the part. In this paper we present some applications over canonical and complex structures showing the capabilities of phased array techniques and associated simulations. Experiments and simulations carried out over complex mock-ups (in terms of entry profile geometry) with embedded reflectors show that these tools allow to efficiently conceive phased array methods and to predict their actual performances. Introduction: Significant technological advances have been made over last years in phased arrays techniques, mostly in terms of acquisition systems (versatility, miniaturization) and transducers technology (piezocomposite probes with highly reduced acoustical cross-talk). Such progress makes the phased array technology seen as a very powerful tool in terms of adaptability and versatility for a wide range of industrial applications. However, as a wide range of complex applications – beam-steering and electronic commutation, multiple parameters settings -, can be performed with such systems, an optimal utilization requires modelingbased conception and exploitation. Modeling is applied to compute suitable delay laws, to simulate the inspection, allowing feasibility study or performance demonstration and reconstruction of acquired data. In this paper, we will briefly present the UT simulations codes developped at the French Atomic Energy Commission (CEA) for several years, then we will present some validation examples as well as simulated applications of them for different phased array inspection techniques. Although conventional techniques may be readily simulated most examples will deal with socalled “dynamic inspection modes”, which postulate that for each scanning position (if any), multiple settings may be applied to the phased array probe. Modeling tools allow to simulate realistic NDE configurations inspections. These models (for ultrasonic as well as eddy current techniques), gathered in the Civa software [1-2], aim at being able to conceive, optimize and predict the performances of various NDE methods. Such models may also be used for experimental data inversion [3] or complex results interpretation. A very broad range of realistic configurations has to be dealt with, in terms of : • Specimen (isotropic or anisotropic, homogeneous or heterogenous, of simple or complex – possibly CAD – geometry) • Probes : standard or advanced, e.g. phased arrays • Scatterers : calibration defects or complex shaped defects, solid inclusions • Inspection method : pulse-echo, TOFD, tandem applications • 3D and broadband regime for realistic echo simulation Such configurations also need to be modeled with high speed computation codes for parametrical studies. Semi-analytical models have therefore been developed : they include the simulation of the beam propagation as well as defect scattering. The computation of the beam radiated by the transducer through the specimen relies on the summation of every elementary contributions over the surface of the probe. Each elementary contribution is evaluated by means of a pencil method [4]. The “pencil” term corresponds to a collection of rays emanating from the point source, which will then diverge during the propagation. The central ray of this pencil lies over a geometrical path between the source point and the computation point, which respects Snell-Descartes ‘s laws for refraction/reflection at any interface. The divergence factor, which takes account of the beam spreading over the propagation is evaluated using the energy conservation over the envelope of the pencil. Once all of the contributions have been evaluated, the wave field transmitted is synthesized and expressed in the pulse response formulation, which enables to predict the acoustic behavior of the probe for any arbitrary waveform. For phased array computations, such a formulation allows to take account of arbitrary delay and amplitude laws applications over the array without any need of new calculations : the pulse responses of all elements of the array are individually computed and stored. Therefore the beam radiated by the array may be readily synthesized as the sum of individual pulse responses, time shifted and weighted according to the delay and amplitude laws [5]. Different defect scattering computations approximations may be used, depending on the type of method and on the defect. As long as void cracks are dealt with, the Kirchhoff approximation is used for the defect scattering computation [6]. This method assumes that the echoes generated by an insonified defect are obtained from the sommation of contributions from each elementary surface of the defect. Those contributions may be seen as scattered wavelets. Their relative amplitudes are calculated using the incident wave field distribution – computed using the pencil method described in the previous paragraph , and the complex diffraction coefficient at the defect. Finally, an argument based on Auld’s reciprocity is used to predict the sensitivity at reception, that-is-to-say the signal observed by the receiving probe. Each possible echo formation, in Longitudinal (L) waves mode, Transverse (T) wave mode or using mode conversion at the backwall or at the defect are individually computed. The overall response obtained for the inspection simulation results from the sommation of these individual echoes. It may also be mentionned that some developments have recently been performed to deal with solid inclusions [7] – for which the Kirchhoff approximation is not valid anymore , as well as complex scattering effects which may occur on non smooth defects (multiple paths over a ramified defect, etc...). For this latter case, an hybrid method was used to couple a finite element code for scattering around the defect and the pencil method carried out for the long range propagation modelling [8]. Results: In the following, we present some examples of comparison between simulations and experiments concerning both the field computation model and the defect scattering model, prior to specific simulations that illustrates some skills of phased array techniques in terms of phase aberration correction abilities, electronic commutation performances assessment and dynamic tandem inspection with phased arrays compared to a classical tandem inspection with a pair of probes. Validation of beam and defect scattering models Figure 1 below shows an example of comparison between simulated and measured beam profiles radiated by a circular arrays used in immersion mode (made of 11 rings, and with a central frequency of 1 MHz). Measured and simulated profiles are located at 20 and 50 mm depth, along aline perpendicular to the probe axis. The transducer radiates longitudinal waves focused at 50 mm depth in a ferritic steel block. Fields profiles measurements are obtained with an electromagnetic probe (EMAT) in through transmission tests over a stepped block. Simulations are carried out with a reference waveform corresponding to the ascan obtained at the focusing point. The use of an experimental signal as the reference waveform allows to take account of the acquisition system characteristics as well as the bandwidth of the receiving probe. The very good agrement between simulations and measurements both validates the model and the manufactured probe. It also allows to show that the electrical and acoustic cross-talk between elements of the array are very low, otherwise they would have led to additional contributions. Profile at 50 mm (focal depth) Measured Simulated